Can Learning Guide Evolution?
• “Baldwin Effect”:
– proposed independently in 1890s by Baldwin, Poulton,
C. Lloyd Morgan
– spread of genetic predispositions to acquire certain
knowledge/skills
• Gene-culture coevolution
• Special case of niche construction: organisms
shape the environments in which they evolve
• Also involves extragenetic inheritance
• Indirect causal paths from individual adaptation to
genome
11/17/03
1
Evolution in Broad Sense
• Evolution in the broadest terms:
– blind variation
– selective retention
• Has been applied to nonbiological evolution
– evolutionary epistemology
– creativity
– memes
11/17/03
2
Genetic Algorithms
• Developed by John Holland in ‘60s
• Did not become popular until late ‘80s
• A simplified model of genetics and
evolution by natural selection
• Most widely applied to optimization
problems (maximize “fitness”)
11/17/03
3
Assumptions
• Existence of fitness function to quantify
merit of potential solutions
– this “fitness” is what the GA will maximize
• A mapping from bit-strings to potential
solutions
– best if each possible string generates a legal
potential solution
– choice of mapping is important
– can use strings over other finite alphabets
11/17/03
4
Outline of Simplified GA
1. Random initial population P(0)
2. Repeat for t = 0, …, tmax or until
converges:
a) create empty population P(t + 1)
b) repeat until P(t + 1) is full:
1)
2)
3)
4)
11/17/03
select two individuals from P(t) based on fitness
optionally mate & replace with offspring
optionally mutate offspring
add two individuals to P(t + 1)
5
Fitness-Biased Selection
• Want the more “fit” to be more likely to
reproduce
– always selecting the best
fi premature convergence
– probabilistic selection fi better exploration
• Roulette-wheel selection: probability µ
relative fitness:
Pr{i mates} =
11/17/03
fi
Â
n
j=1
fj
6
Crossover: Biological Inspiration
• Occurs during
meiosis, when haploid
gametes are formed
• Randomly mixes
genes from two
parents
• Creates genetic
variation in gametes
11/17/03
(fig. from B&N Thes. Biol.)
7
GAs: One-point Crossover
parents
11/17/03
offspring
8
GAs: Two-point Crossover
parents
11/17/03
offspring
9
GAs: N-point Crossover
parents
11/17/03
offspring
10
Mutation: Biological Inspiration
• Chromosome mutation fi
• Gene mutation: alteration
of the DNA in a gene
– inspiration for mutation in
GAs
• In typical GA each bit has
a low probability of
changing
• Some GAs models
rearrange bits
11/17/03
(fig. from B&N Thes. Biol.)
11
Example: GA for IPD
• Genetic strings encode strategy
– for first round
– based on self’s & opponent’s action on r
previous rounds
– hence 22r + 1 bits
• E.g., for r = 1:
opp. cooperated
opp. defected
we cooperated
first round
opp. defected
opp. cooperated
11/17/03
we defected
12
Typical Result
convergence
11/17/03
13
The Red Queen Hypothesis
“Now, here, you see, it takes
all the running you can do,
to keep in the same place.”
— Through the Looking-Glass
and What Alice Found There
11/17/03
• Observation: a species
probability of extinction is independent of
time it has existed
• Hypothesis: species
continually adapt to
each other
• Extinction occurs with
insufficient variability
for further adaptation
14